OpenAI Codex
- Surfaces represented
- Codex CLI
- Model context represented
- GPT-5.2
Practical agent comparison
Codex vs Gemini CLI
Both terminal agents document project guidance, planning, tools, file modification, shell execution, and approval boundaries. Codex uses AGENTS.md, and its prompt claims are scoped to Codex CLI with GPT-5.2. Gemini CLI uses hierarchical GEMINI.md context with a configurable model. Both document web functionality, while Codex CLI distinguishes cached and live search. Recheck context assembly, network needs, and execution controls before moving a workflow.
- Reviewed
- Products
- OpenAI Codex · Gemini CLI
- Decision criteria
- 9 reviewed questions
Comparison orientation
Four reviewed dimensions at a glance
Each dimension links to its detailed evidence row. State labels describe the selected evidence.
Left productOpenAI CodexOpen product record
- Documented 01Repository / project instructionsOpen detailed evidence Documented
- Not established 02Memory / persistent contextOpen detailed evidence Not established
- Documented 03Planning / work decompositionOpen detailed evidence Documented
- Documented 04Tools / tool callingOpen detailed evidence Documented
Who this comparison is for
How do Codex and Gemini CLI differ for professional terminal coding workflows?
Developers and platform teams choosing or migrating between Codex CLI and Gemini CLI for repository-centered terminal work.
Evidence-bounded scope
Product, surface, and model context
These summaries combine only the scope stated in the reviewed answers. Each criterion keeps its narrower limits.
Gemini CLI
- Surfaces represented
- Gemini CLI
- Model context represented
- Not established
Practical behavior comparison
What the selected evidence establishes
“Not established” records an evidence gap. It does not mean the product lacks the capability.
| Criterion | OpenAI Codex | Gemini CLI | Interpretation |
|---|---|---|---|
| Repository / project instructions Repository guidance must be placed in the target agent's documented instruction hierarchy. | Documented AGENTS.md repository guidance
| Documented Hierarchical GEMINI.md context
| Read practical implication |
| Memory / persistent context Durable workflow context should not be assumed to persist across agent environments. | Not established Persistent memory not established
| Not established Persistent memory semantics not established
| Read practical implication |
| Planning / work decomposition Planning assumptions shape how teams structure, review, and resume multi-step work. | Documented Making and updating plans
| Documented Planning appears in documented tools
| Read practical implication |
| Tools / tool calling A portable workflow must use only the tools established by the selected evidence. | Documented Terminal and patch function calls
| Documented Filesystem, execution, planning, MCP, and web
| Read practical implication |
| Filesystem / file modification Repository reads and mutations require the correct workspace and review boundary. | Documented Workspace files and patch calls
| Documented Filesystem tools with confirmation for mutation
| Read practical implication |
| Shell / terminal execution Command-heavy automation depends on the target harness's execution controls. | Documented Terminal command function calls
| Documented Shell tools with confirmation and isolation
| Read practical implication |
| Browsing / network access Current external information cannot be assumed available when browsing scope is unestablished. | Documented Cached or live CLI web search
| Documented Web tools in official tool reference
| Read practical implication |
| Approvals / permissions Consent, confirmation, and isolation rules determine which delegated actions can proceed. | Documented Configuration-dependent approval escalation
| Documented Confirmation, sandbox, and trusted folders
| Read practical implication |
| Context scope / inheritance Instructions and tools remain dependable only when their surface, model, and configuration boundaries are preserved. | Documented Harness, workspace, and repository context
| Documented Global, workspace, just-in-time, and dynamic context
| Read practical implication |
Criterion-by-criterion interpretation
Scope, limitations, and evidence
Behavior Intelligence supplies the factual answers. Editorial implications remain separate.
01
Open evidence across six products How do AI agents use repository or project instructions?
Why it matters. Repository guidance must be placed in the target agent's documented instruction hierarchy.
OpenAI Codex
DocumentedCodex supports repository-scoped guidance through AGENTS.md and AGENTS.override.md in the documented context assembly. Codex home guidance loads before project guidance; project discovery selects at most one file per directory from the project root to the working directory.
- Surface
- Codex CLI
- Model context
- Not established
- Configuration / version
- Instruction discovery follows the documented repository path and size limits.
Limits and uncertainty
- The selected evidence does not establish every runtime instruction input.
Gemini CLI
DocumentedGemini CLI documents global, workspace, and just-in-time GEMINI.md files as hierarchical project and task context.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Not established
Limits and uncertainty
- The context hierarchy is distinct from the replaceable built-in system prompt.
02
Open evidence across six products Does the product document memory or persistent context?
Why it matters. Durable workflow context should not be assumed to persist across agent environments.
OpenAI Codex
Not establishedThe limited Codex excerpt and repository-guidance source do not establish a separate persistent memory mechanism.
- Surface
- Not established
- Model context
- Not established
- Configuration / version
- Not established
Limits and uncertainty
- Workspace context and AGENTS.md instructions are not treated as memory.
No source link is attached because the reviewed evidence does not establish this answer.
Gemini CLI
Not establishedThe selected Gemini CLI tools reference names memory among its categories, but does not establish enough persistence or scope behavior for this question.
- Surface
- Not established
- Model context
- Not established
- Configuration / version
- Not established
Limits and uncertainty
- A tool category alone is not treated as proof of cross-session persistence.
No source link is attached because the reviewed evidence does not establish this answer.
03
Open evidence across six products How is planning or work decomposition documented?
Why it matters. Planning assumptions shape how teams structure, review, and resume multi-step work.
OpenAI Codex
DocumentedThe selected Codex prompt excerpt explicitly names making and updating plans as part of the assistant's working interface.
- Surface
- Codex CLI
- Model context
- GPT-5.2
- Configuration / version
- Not established
Limits and uncertainty
- The excerpt does not prescribe one universal decomposition procedure.
Gemini CLI
DocumentedThe selected Gemini CLI tools reference includes planning among the documented tool categories available to the CLI.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Not established
Limits and uncertainty
- The selected evidence does not establish one fixed planning sequence across configurations.
04
Open evidence across six products What tools or tool-calling capabilities are documented?
Why it matters. A portable workflow must use only the tools established by the selected evidence.
OpenAI Codex
DocumentedCodex's selected prompt excerpt names function calls for terminal commands and patch application.
- Surface
- Codex CLI
- Model context
- GPT-5.2
- Configuration / version
- Actual tool availability and escalation depend on the specific run configuration.
Limits and uncertainty
- The excerpt is not a complete tool inventory.
Gemini CLI
DocumentedGemini CLI's selected official reference covers filesystem, execution, planning, memory, MCP, and web tool categories.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Enabled tools and security posture can vary with configuration.
Limits and uncertainty
- The selected evidence does not establish one immutable tool set.
05
Open evidence across six products Can the agent read or modify files?
Why it matters. Repository reads and mutations require the correct workspace and review boundary.
OpenAI Codex
DocumentedCodex's selected excerpt says the harness can supply workspace files and that the assistant can emit patch function calls.
- Surface
- Codex CLI
- Model context
- GPT-5.2
- Configuration / version
- Workspace access and patch approval depend on the run configuration.
Limits and uncertainty
- The selected excerpt does not establish unrestricted filesystem access.
Gemini CLI
DocumentedGemini CLI documents filesystem tools, with mutating file operations subject to confirmation and configured security boundaries.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Sandbox and trusted-folder settings can change the execution boundary.
Limits and uncertainty
- Tool availability varies by configuration and version.
06
Open evidence across six products Can the agent execute shell or terminal commands?
Why it matters. Command-heavy automation depends on the target harness's execution controls.
OpenAI Codex
DocumentedCodex's selected excerpt explicitly names function calls that run terminal commands.
- Surface
- Codex CLI
- Model context
- GPT-5.2
- Configuration / version
- Commands may require user approval depending on the run configuration.
Limits and uncertainty
- The selected excerpt does not establish unrestricted command execution.
Gemini CLI
DocumentedGemini CLI documents shell execution tools whose requests pass through confirmation and configured security policies.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Sandboxing and trusted-folder settings can constrain execution.
Limits and uncertainty
- The enabled tool set is configuration-dependent.
07
Open evidence across six products Can the product browse or access web and network resources?
Why it matters. Current external information cannot be assumed available when browsing scope is unestablished.
OpenAI Codex
DocumentedInteractive Codex CLI defaults to cached web search; --search selects live web search. This tool capability does not establish unrestricted shell network access or every session configuration.
- Surface
- Codex CLI
- Model context
- Not established
- Configuration / version
- Current product documentation consulted 2026-09-14; enabled mode, settings, and deployment still determine availability.
Limits and uncertainty
- This claim does not update the dated prompt snapshot or establish every runtime configuration.
Gemini CLI
DocumentedGemini CLI's selected official tools reference includes web capabilities among the available tool categories.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Availability depends on enabled tools and security configuration.
Limits and uncertainty
- The selected evidence does not establish unrestricted network access.
08
Open evidence across six products What approval or permission controls are documented?
Why it matters. Consent, confirmation, and isolation rules determine which delegated actions can proceed.
OpenAI Codex
DocumentedCodex's selected excerpt says terminal and patch calls may be escalated to the user for approval depending on configuration.
- Surface
- Codex CLI
- Model context
- GPT-5.2
- Configuration / version
- The specific run determines which calls require escalation.
Limits and uncertainty
- The excerpt does not define every sandbox or approval mode.
Gemini CLI
DocumentedGemini CLI documents confirmation for mutating file and shell tools, with sandboxing and trusted-folder boundaries.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- Security policy and isolation settings shape the actual boundary.
Limits and uncertainty
- The selected evidence does not establish every enterprise policy.
09
Open evidence across six products How is context scoped or inherited across products and agents?
Why it matters. Instructions and tools remain dependable only when their surface, model, and configuration boundaries are preserved.
OpenAI Codex
DocumentedCodex's selected sources establish harness-supplied workspace context and authored guidance from Codex home followed by project instruction files. Current discovery documentation is separate from the pinned GPT-5.2 prompt excerpt.
- Surface
- Codex CLI
- Model context
- Not established
- Configuration / version
- The current instruction-discovery claim covers Codex CLI; the historical harness excerpt remains scoped to GPT-5.2.
Limits and uncertainty
- The complete runtime context assembly is not established.
Gemini CLI
DocumentedGemini CLI documents global, workspace, and just-in-time GEMINI.md context plus dynamic injection of enabled skills, subagents, and tools.
- Surface
- Gemini CLI
- Model context
- Not established
- Configuration / version
- The product uses a configurable Gemini model and can fully replace its built-in system prompt.
Limits and uncertainty
- No single immutable runtime assembly is claimed.
Workflow guidance
Migration considerations
These implications follow from the reviewed criteria. They do not express a product preference.
Example: preserve nested scope in a two-service repository
Transfer a narrow service exception without letting it become a global instruction.
- Suppose repository AGENTS.md names the normal test command, services/payments/AGENTS.override.md supplies a payment-specific command, and Codex home AGENTS.md holds a personal output preference. Start Codex from services/payments when evaluating that documented discovery chain.
- Map the shared team guidance into workspace GEMINI.md and the personal preference into global GEMINI.md context. Preserve the payment exception only where Gemini’s documented contextual discovery applies to that service; do not place it in the global file.
- AGENTS.override.md selects a file at one Codex directory level. Do not assume renaming it to GEMINI.md preserves that replacement behavior: inspect the resulting global, workspace, and just-in-time context and resolve conflicting commands explicitly.
- For a task that checks current API documentation before editing, verify live versus cached Codex search and the enabled Gemini web tools. Ask for the checked source URL and its version, then review the plan before permitting repository mutation.
- Handoff condition: a payment task receives the payment command while another service retains the shared command. Confirm each destination context and its shell controls separately. Keep plan and progress notes explicit; neither selected record establishes comparable automatic memory.
Keep planning and state explicit
The reviewed evidence establishes planning for both products, but not comparable persistent memory. Keep plan and state artifacts outside assumed memory.
Preserve execution controls
Filesystem and shell automation must retain Codex escalation or Gemini CLI confirmation and isolation rather than assuming equivalent unattended execution.
Audit network dependencies
For current release information, select live search in Codex CLI when available and inspect the Gemini CLI web-tool configuration. Require cited page evidence in both; shell network permission remains a separate check.
Comparison-specific evidence
Differences in available evidence
First-party evidence covers project context, planning, tools, file changes, command execution, and approval boundaries for both terminal coding agents. Their prompt and configuration scopes remain distinct. The analysis below adds prompt and product context to the shared behavior criteria.
- 01
The available prompt evidence has different boundaries
Codex provides a limited pinned prompt excerpt. Gemini CLI documents built-in prompt export and full replacement. Both retain completeness and configuration limits.
- 02
Both are terminal agent products
Each product harness remains distinct from its selected or configurable model context.
Source-specific detail
Product and prompt evidence
These criteria cover prompt visibility, product identity, and other evidence specific to this comparison.
| Criterion | OpenAI Codex | Gemini CLI |
|---|---|---|
| Product and model roleEditorial interpretation. Both are terminal coding agents whose model context is separately scoped. | documented Codex CLI is the represented product surface; GPT-5.2 is the model named by the selected pinned prompt rather than every Codex runtime. | documented Gemini CLI is the represented terminal product, with a configurable Gemini model kept separate from instruction assembly. |
| Prompt evidence postureEditorial interpretation. The available prompt artifacts differ, and each retains its own completeness and configuration limits. | documented Promptcred maps a limited excerpt from a pinned first-party Codex prompt file; it does not claim the complete runtime assembly. | documented Gemini CLI documentation establishes built-in prompt export and complete replacement workflows whose resolved content can depend on configuration. |
Using the evidence
How to read this evidence
Keep product and model scope separate
Codex CLI with the selected GPT-5.2 prompt and Gemini CLI with a configurable Gemini model are bounded evidence, not universal model claims.
Use each artifact according to its publication posture
A limited pinned excerpt and a documentation-defined export or replacement workflow support different prompt-level conclusions.
Primary-source trail
Sources used in this comparison
Each source links to its product evidence, with provenance and publication limits preserved.
OpenAI Codex
- Tier A · vendor-publishedCodex custom instructions with AGENTS.md
- Tier A · vendor-publishedCodex CLI command reference — web search
- Tier B · first-party open sourceOpenAI Codex prompt file
- Tier A · vendor-publishedUnrolling the Codex agent loop
Gemini CLI
- Tier A · vendor-publishedGemini CLI system-prompt override documentation
- Tier A · vendor-publishedGemini CLI GEMINI.md context documentation
- Tier A · vendor-publishedGemini CLI tools reference
Deeper cross-agent context
Behavior questions in this comparison
Open the six-product answer for broader evidence and scope.
Repository / project instructionsHow do AI agents use repository or project instructions?Memory / persistent contextDoes the product document memory or persistent context?Planning / work decompositionHow is planning or work decomposition documented?Tools / tool callingWhat tools or tool-calling capabilities are documented?Filesystem / file modificationCan the agent read or modify files?Shell / terminal executionCan the agent execute shell or terminal commands?Browsing / network accessCan the product browse or access web and network resources?Approvals / permissionsWhat approval or permission controls are documented?Context scope / inheritanceHow is context scoped or inherited across products and agents?
Continue comparing